Face Generation

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data

You'll be using two datasets in this project:

  • MNIST
  • CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

In [1]:
# data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)
Found mnist Data
Found celeba Data

Explore the Data

MNIST

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

In [2]:
show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')
Out[2]:
<matplotlib.image.AxesImage at 0x7fe8631d0400>

CelebA

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

In [3]:
show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))
Out[3]:
<matplotlib.image.AxesImage at 0x7fe8630d24e0>

Preprocess the Data

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network

You'll build the components necessary to build a GANs by implementing the following functions below:

  • model_inputs
  • discriminator
  • generator
  • model_loss
  • model_opt
  • train

Check the Version of TensorFlow and Access to GPU

This will check to make sure you have the correct version of TensorFlow and access to a GPU

In [4]:
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))
TensorFlow Version: 1.1.0
Default GPU Device: /gpu:0

Input

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

  • Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
  • Z input placeholder with rank 2 using z_dim.
  • Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

In [7]:
import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    images = tf.placeholder(tf.float32, shape=(None, image_width, image_height, image_channels), name="images_real")
    z = tf.placeholder(tf.float32, shape=(None, z_dim), name="latent")
    learning_rate = tf.placeholder(tf.float32, name="learning_rate")

    return images, z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)
Tests Passed

Discriminator

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the generator, tensor logits of the generator).

In [8]:
def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param image: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    
    
    alpha = 0.2
    with tf.variable_scope('discriminator', reuse=reuse):
        # Input layer is 28x28x3
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same')
        relu1 = tf.maximum(alpha * x1, x1)
        # 14x14x64
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same')
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        # 7x7x128
        
        x3 = tf.layers.conv2d(relu2, 256, 5, strides=2, padding='same')
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
        # 4x4x256

        # Flatten it
        flat = tf.reshape(relu3, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)
Tests Passed

Generator

Implement generator to generate an image using z. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

In [19]:
def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    
    alpha = 0.2
    with tf.variable_scope('generator', reuse=(not is_train)):
        # First fully connected layer
        x1 = tf.layers.dense(z, 4*4*512)
        
        # Reshape it to start the convolutional stack
        x1 = tf.reshape(x1, (-1, 4, 4, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        # 4x4x512
        
        x3 = tf.layers.conv2d_transpose(x1, 256, 4, strides=1, padding='valid')
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        # 7x7x256
        
        x4 = tf.layers.conv2d_transpose(x3, 256, 5, strides=1, padding='same')
        x4 = tf.layers.batch_normalization(x4, training=is_train)
        x4 = tf.maximum(alpha * x4, x4)
        # 7x7x256
        
        x5 = tf.layers.conv2d_transpose(x4, 128, 5, strides=2, padding='same')
        x5 = tf.layers.batch_normalization(x5, training=is_train)
        x5 = tf.maximum(alpha * x5, x5)
        # 14x14x128
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x5, out_channel_dim, 5, strides=2, padding='same')
        # 28x28x3
        
        out = tf.tanh(logits)
        return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)
Tests Passed

Loss

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

  • discriminator(images, reuse=False)
  • generator(z, out_channel_dim, is_train=True)
In [10]:
def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    images_fake = generator(input_z, out_channel_dim, is_train=True)
    _, d_logits_real = discriminator(input_real, reuse=False)
    _, d_logits_fake = discriminator(images_fake, reuse=True)
    
    # Calculate losses
    smooth = 0.1
    d_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_logits_real)*(1 - smooth)))
    d_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_logits_fake)))
    d_loss = d_loss_real + d_loss_fake
    
    g_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_logits_fake)))
    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)
Tests Passed

Optimization

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

In [21]:
def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # Get variables
    t_vars = tf.trainable_variables()
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]

    # Define optimization operations
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate=learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate=learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)
    
        return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)
Tests Passed

Neural Network Training

Show Output

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

In [22]:
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train

Implement train to build and train the GANs. Use the following functions you implemented:

  • model_inputs(image_width, image_height, image_channels, z_dim)
  • model_loss(input_real, input_z, out_channel_dim)
  • model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

In [25]:
def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    _, image_width, image_height, image_channels = data_shape
    images_real, z, lr = model_inputs(image_width, image_height, image_channels, z_dim)
    d_loss, g_loss = model_loss(images_real, z, image_channels)
    d_train_opt, g_train_opt = model_opt(d_loss, g_loss, lr, beta1)
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        i = 0
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                batch_images *= 2
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                _ = sess.run(d_train_opt, feed_dict={images_real: batch_images, z: batch_z, lr: learning_rate})
                _ = sess.run(g_train_opt, feed_dict={images_real: batch_images, z: batch_z, lr: learning_rate})
                if i % 30 == 0:
                    show_generator_output(sess, 25, z, image_channels, data_image_mode)
                    train_loss_d = sess.run(d_loss, {z: batch_z, images_real: batch_images})
                    train_loss_g = g_loss.eval({z: batch_z})
                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                          "Batch {}...".format(i),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                i += 1
                
                

MNIST

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

In [26]:
batch_size = 64
z_dim = 100
learning_rate = 0.005
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)
Epoch 1/2... Batch 0... Discriminator Loss: 37.5859... Generator Loss: 0.0000
Epoch 1/2... Batch 30... Discriminator Loss: 1.4914... Generator Loss: 0.6290
Epoch 1/2... Batch 60... Discriminator Loss: 0.9913... Generator Loss: 0.9242
Epoch 1/2... Batch 90... Discriminator Loss: 1.8101... Generator Loss: 0.4458
Epoch 1/2... Batch 120... Discriminator Loss: 1.8520... Generator Loss: 0.3071
Epoch 1/2... Batch 150... Discriminator Loss: 2.5835... Generator Loss: 3.1513
Epoch 1/2... Batch 180... Discriminator Loss: 1.4761... Generator Loss: 1.4820
Epoch 1/2... Batch 210... Discriminator Loss: 1.2375... Generator Loss: 1.4035
Epoch 1/2... Batch 240... Discriminator Loss: 1.0570... Generator Loss: 1.6272
Epoch 1/2... Batch 270... Discriminator Loss: 1.0689... Generator Loss: 1.0822
Epoch 1/2... Batch 300... Discriminator Loss: 1.3314... Generator Loss: 0.5847
Epoch 1/2... Batch 330... Discriminator Loss: 1.2862... Generator Loss: 1.3384
Epoch 1/2... Batch 360... Discriminator Loss: 1.3183... Generator Loss: 1.0547
Epoch 1/2... Batch 390... Discriminator Loss: 0.8432... Generator Loss: 1.1861
Epoch 1/2... Batch 420... Discriminator Loss: 1.2062... Generator Loss: 1.0336
Epoch 1/2... Batch 450... Discriminator Loss: 1.4391... Generator Loss: 0.5595
Epoch 1/2... Batch 480... Discriminator Loss: 1.6163... Generator Loss: 1.6707
Epoch 1/2... Batch 510... Discriminator Loss: 1.3010... Generator Loss: 1.1163
Epoch 1/2... Batch 540... Discriminator Loss: 1.3691... Generator Loss: 0.6747
Epoch 1/2... Batch 570... Discriminator Loss: 1.4203... Generator Loss: 0.5017
Epoch 1/2... Batch 600... Discriminator Loss: 1.4797... Generator Loss: 0.5833
Epoch 1/2... Batch 630... Discriminator Loss: 1.4426... Generator Loss: 0.5929
Epoch 1/2... Batch 660... Discriminator Loss: 1.1501... Generator Loss: 1.1054
Epoch 1/2... Batch 690... Discriminator Loss: 1.5356... Generator Loss: 1.5864
Epoch 1/2... Batch 720... Discriminator Loss: 1.1978... Generator Loss: 0.7870
Epoch 1/2... Batch 750... Discriminator Loss: 1.3848... Generator Loss: 0.5641
Epoch 1/2... Batch 780... Discriminator Loss: 1.3207... Generator Loss: 0.6454
Epoch 1/2... Batch 810... Discriminator Loss: 1.2206... Generator Loss: 0.9398
Epoch 1/2... Batch 840... Discriminator Loss: 1.2347... Generator Loss: 1.0752
Epoch 1/2... Batch 870... Discriminator Loss: 1.2299... Generator Loss: 1.1958
Epoch 1/2... Batch 900... Discriminator Loss: 1.5078... Generator Loss: 0.4526
Epoch 1/2... Batch 930... Discriminator Loss: 1.3552... Generator Loss: 0.5438
Epoch 2/2... Batch 960... Discriminator Loss: 1.2396... Generator Loss: 1.2001
Epoch 2/2... Batch 990... Discriminator Loss: 1.3815... Generator Loss: 0.5973
Epoch 2/2... Batch 1020... Discriminator Loss: 1.2316... Generator Loss: 1.5614
Epoch 2/2... Batch 1050... Discriminator Loss: 1.2470... Generator Loss: 0.7021
Epoch 2/2... Batch 1080... Discriminator Loss: 1.4660... Generator Loss: 0.5078
Epoch 2/2... Batch 1110... Discriminator Loss: 1.1736... Generator Loss: 0.8687
Epoch 2/2... Batch 1140... Discriminator Loss: 1.4099... Generator Loss: 0.5010
Epoch 2/2... Batch 1170... Discriminator Loss: 1.4369... Generator Loss: 0.5417
Epoch 2/2... Batch 1200... Discriminator Loss: 1.0414... Generator Loss: 1.2894
Epoch 2/2... Batch 1230... Discriminator Loss: 1.1873... Generator Loss: 1.0803
Epoch 2/2... Batch 1260... Discriminator Loss: 1.4777... Generator Loss: 0.4967
Epoch 2/2... Batch 1290... Discriminator Loss: 1.2802... Generator Loss: 0.6169
Epoch 2/2... Batch 1320... Discriminator Loss: 1.2574... Generator Loss: 1.2521
Epoch 2/2... Batch 1350... Discriminator Loss: 1.3819... Generator Loss: 0.5998
Epoch 2/2... Batch 1380... Discriminator Loss: 1.1024... Generator Loss: 0.8691
Epoch 2/2... Batch 1410... Discriminator Loss: 1.1453... Generator Loss: 0.9465
Epoch 2/2... Batch 1440... Discriminator Loss: 1.0455... Generator Loss: 1.5108
Epoch 2/2... Batch 1470... Discriminator Loss: 1.3812... Generator Loss: 1.2446
Epoch 2/2... Batch 1500... Discriminator Loss: 1.5163... Generator Loss: 1.9461
Epoch 2/2... Batch 1530... Discriminator Loss: 1.3529... Generator Loss: 0.5457
Epoch 2/2... Batch 1560... Discriminator Loss: 0.9043... Generator Loss: 1.1049
Epoch 2/2... Batch 1590... Discriminator Loss: 0.8925... Generator Loss: 1.6364
Epoch 2/2... Batch 1620... Discriminator Loss: 1.1445... Generator Loss: 1.3524
Epoch 2/2... Batch 1650... Discriminator Loss: 1.7540... Generator Loss: 1.9698
Epoch 2/2... Batch 1680... Discriminator Loss: 1.6990... Generator Loss: 0.4130
Epoch 2/2... Batch 1710... Discriminator Loss: 0.8428... Generator Loss: 1.1562
Epoch 2/2... Batch 1740... Discriminator Loss: 1.0821... Generator Loss: 1.0125
Epoch 2/2... Batch 1770... Discriminator Loss: 0.8733... Generator Loss: 1.0901
Epoch 2/2... Batch 1800... Discriminator Loss: 1.3005... Generator Loss: 0.6010
Epoch 2/2... Batch 1830... Discriminator Loss: 0.9872... Generator Loss: 1.2701
Epoch 2/2... Batch 1860... Discriminator Loss: 1.4055... Generator Loss: 0.5107

CelebA

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

In [27]:
batch_size = 64
z_dim = 100
learning_rate = 0.005
beta1 = 0.1

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)
Epoch 1/1... Batch 0... Discriminator Loss: 20.8165... Generator Loss: 0.0000
Epoch 1/1... Batch 30... Discriminator Loss: 2.0090... Generator Loss: 0.9807
Epoch 1/1... Batch 60... Discriminator Loss: 1.3336... Generator Loss: 2.2151
Epoch 1/1... Batch 90... Discriminator Loss: 2.4552... Generator Loss: 1.2781
Epoch 1/1... Batch 120... Discriminator Loss: 0.7668... Generator Loss: 2.7716
Epoch 1/1... Batch 150... Discriminator Loss: 0.6296... Generator Loss: 2.1472
Epoch 1/1... Batch 180... Discriminator Loss: 0.6917... Generator Loss: 1.3427
Epoch 1/1... Batch 210... Discriminator Loss: 0.5928... Generator Loss: 1.7557
Epoch 1/1... Batch 240... Discriminator Loss: 0.5529... Generator Loss: 2.8059
Epoch 1/1... Batch 270... Discriminator Loss: 0.4010... Generator Loss: 3.2249
Epoch 1/1... Batch 300... Discriminator Loss: 1.3873... Generator Loss: 1.2839
Epoch 1/1... Batch 330... Discriminator Loss: 4.0794... Generator Loss: 4.5179
Epoch 1/1... Batch 360... Discriminator Loss: 0.9789... Generator Loss: 0.7759
Epoch 1/1... Batch 390... Discriminator Loss: 0.4233... Generator Loss: 2.7407
Epoch 1/1... Batch 420... Discriminator Loss: 2.2175... Generator Loss: 2.7197
Epoch 1/1... Batch 450... Discriminator Loss: 1.6686... Generator Loss: 0.6702
Epoch 1/1... Batch 480... Discriminator Loss: 0.6526... Generator Loss: 1.4178
Epoch 1/1... Batch 510... Discriminator Loss: 1.3123... Generator Loss: 0.6564
Epoch 1/1... Batch 540... Discriminator Loss: 1.4257... Generator Loss: 0.4848
Epoch 1/1... Batch 570... Discriminator Loss: 1.2052... Generator Loss: 0.7563
Epoch 1/1... Batch 600... Discriminator Loss: 1.2064... Generator Loss: 0.9201
Epoch 1/1... Batch 630... Discriminator Loss: 1.6177... Generator Loss: 0.4259
Epoch 1/1... Batch 660... Discriminator Loss: 1.3986... Generator Loss: 0.6823
Epoch 1/1... Batch 690... Discriminator Loss: 1.3048... Generator Loss: 0.9563
Epoch 1/1... Batch 720... Discriminator Loss: 1.2645... Generator Loss: 0.7586
Epoch 1/1... Batch 750... Discriminator Loss: 1.3299... Generator Loss: 0.8737
Epoch 1/1... Batch 780... Discriminator Loss: 1.3148... Generator Loss: 0.6969
Epoch 1/1... Batch 810... Discriminator Loss: 1.3207... Generator Loss: 0.6197
Epoch 1/1... Batch 840... Discriminator Loss: 1.2324... Generator Loss: 0.9507
Epoch 1/1... Batch 870... Discriminator Loss: 1.4669... Generator Loss: 0.4656
Epoch 1/1... Batch 900... Discriminator Loss: 1.3020... Generator Loss: 0.5531
Epoch 1/1... Batch 930... Discriminator Loss: 1.2601... Generator Loss: 1.0708
Epoch 1/1... Batch 960... Discriminator Loss: 1.3138... Generator Loss: 0.9890
Epoch 1/1... Batch 990... Discriminator Loss: 1.5329... Generator Loss: 0.5204
Epoch 1/1... Batch 1020... Discriminator Loss: 1.3733... Generator Loss: 0.5501
Epoch 1/1... Batch 1050... Discriminator Loss: 1.2178... Generator Loss: 1.2196
Epoch 1/1... Batch 1080... Discriminator Loss: 1.1745... Generator Loss: 1.1835
Epoch 1/1... Batch 1110... Discriminator Loss: 1.2947... Generator Loss: 0.6587
Epoch 1/1... Batch 1140... Discriminator Loss: 1.7354... Generator Loss: 0.3968
Epoch 1/1... Batch 1170... Discriminator Loss: 0.7255... Generator Loss: 1.2431
Epoch 1/1... Batch 1200... Discriminator Loss: 1.4243... Generator Loss: 0.5505
Epoch 1/1... Batch 1230... Discriminator Loss: 1.3471... Generator Loss: 0.5775
Epoch 1/1... Batch 1260... Discriminator Loss: 1.2520... Generator Loss: 0.8197
Epoch 1/1... Batch 1290... Discriminator Loss: 1.3372... Generator Loss: 0.8341
Epoch 1/1... Batch 1320... Discriminator Loss: 1.3221... Generator Loss: 0.8250
Epoch 1/1... Batch 1350... Discriminator Loss: 1.3167... Generator Loss: 1.0094
Epoch 1/1... Batch 1380... Discriminator Loss: 1.4145... Generator Loss: 1.2051
Epoch 1/1... Batch 1410... Discriminator Loss: 1.5453... Generator Loss: 1.2867
Epoch 1/1... Batch 1440... Discriminator Loss: 1.3787... Generator Loss: 0.9396
Epoch 1/1... Batch 1470... Discriminator Loss: 1.3478... Generator Loss: 0.5658
Epoch 1/1... Batch 1500... Discriminator Loss: 1.3220... Generator Loss: 1.1221
Epoch 1/1... Batch 1530... Discriminator Loss: 1.6679... Generator Loss: 0.6753
Epoch 1/1... Batch 1560... Discriminator Loss: 1.2774... Generator Loss: 0.9292
Epoch 1/1... Batch 1590... Discriminator Loss: 1.0755... Generator Loss: 1.1006
Epoch 1/1... Batch 1620... Discriminator Loss: 1.3770... Generator Loss: 0.5522
Epoch 1/1... Batch 1650... Discriminator Loss: 2.2331... Generator Loss: 2.3592
Epoch 1/1... Batch 1680... Discriminator Loss: 1.3073... Generator Loss: 0.7213
Epoch 1/1... Batch 1710... Discriminator Loss: 1.2905... Generator Loss: 0.7217
Epoch 1/1... Batch 1740... Discriminator Loss: 0.7682... Generator Loss: 1.1590
Epoch 1/1... Batch 1770... Discriminator Loss: 1.3432... Generator Loss: 0.7407
Epoch 1/1... Batch 1800... Discriminator Loss: 1.4974... Generator Loss: 0.5725
Epoch 1/1... Batch 1830... Discriminator Loss: 1.5808... Generator Loss: 0.4573
Epoch 1/1... Batch 1860... Discriminator Loss: 0.9176... Generator Loss: 0.9875
Epoch 1/1... Batch 1890... Discriminator Loss: 0.8589... Generator Loss: 0.9949
Epoch 1/1... Batch 1920... Discriminator Loss: 1.4077... Generator Loss: 0.9636
Epoch 1/1... Batch 1950... Discriminator Loss: 1.4763... Generator Loss: 0.7807
Epoch 1/1... Batch 1980... Discriminator Loss: 1.1854... Generator Loss: 0.9137
Epoch 1/1... Batch 2010... Discriminator Loss: 1.2940... Generator Loss: 0.6800
Epoch 1/1... Batch 2040... Discriminator Loss: 1.2410... Generator Loss: 0.7480
Epoch 1/1... Batch 2070... Discriminator Loss: 1.3132... Generator Loss: 1.3613
Epoch 1/1... Batch 2100... Discriminator Loss: 1.2044... Generator Loss: 0.9563
Epoch 1/1... Batch 2130... Discriminator Loss: 1.3137... Generator Loss: 0.7969
Epoch 1/1... Batch 2160... Discriminator Loss: 1.4245... Generator Loss: 0.5187
Epoch 1/1... Batch 2190... Discriminator Loss: 1.9179... Generator Loss: 1.0503
Epoch 1/1... Batch 2220... Discriminator Loss: 1.3155... Generator Loss: 0.6876
Epoch 1/1... Batch 2250... Discriminator Loss: 1.3402... Generator Loss: 1.1632
Epoch 1/1... Batch 2280... Discriminator Loss: 1.3612... Generator Loss: 0.6314
Epoch 1/1... Batch 2310... Discriminator Loss: 1.2584... Generator Loss: 0.9186
Epoch 1/1... Batch 2340... Discriminator Loss: 1.3895... Generator Loss: 1.1526
Epoch 1/1... Batch 2370... Discriminator Loss: 1.3032... Generator Loss: 0.7685
Epoch 1/1... Batch 2400... Discriminator Loss: 1.4016... Generator Loss: 0.5396
Epoch 1/1... Batch 2430... Discriminator Loss: 1.3722... Generator Loss: 0.5898
Epoch 1/1... Batch 2460... Discriminator Loss: 0.9912... Generator Loss: 0.8591
Epoch 1/1... Batch 2490... Discriminator Loss: 1.7217... Generator Loss: 2.0553
Epoch 1/1... Batch 2520... Discriminator Loss: 1.3020... Generator Loss: 0.7464
Epoch 1/1... Batch 2550... Discriminator Loss: 1.2752... Generator Loss: 1.0376
Epoch 1/1... Batch 2580... Discriminator Loss: 1.4270... Generator Loss: 0.5938
Epoch 1/1... Batch 2610... Discriminator Loss: 1.2570... Generator Loss: 0.6857
Epoch 1/1... Batch 2640... Discriminator Loss: 0.4649... Generator Loss: 2.3123
Epoch 1/1... Batch 2670... Discriminator Loss: 1.3810... Generator Loss: 0.6385
Epoch 1/1... Batch 2700... Discriminator Loss: 1.3954... Generator Loss: 0.6421
Epoch 1/1... Batch 2730... Discriminator Loss: 1.3702... Generator Loss: 0.6702
Epoch 1/1... Batch 2760... Discriminator Loss: 1.3452... Generator Loss: 0.8126
Epoch 1/1... Batch 2790... Discriminator Loss: 1.3228... Generator Loss: 0.8722
Epoch 1/1... Batch 2820... Discriminator Loss: 1.3045... Generator Loss: 0.7857
Epoch 1/1... Batch 2850... Discriminator Loss: 1.2907... Generator Loss: 1.2160
Epoch 1/1... Batch 2880... Discriminator Loss: 1.3460... Generator Loss: 0.7059
Epoch 1/1... Batch 2910... Discriminator Loss: 1.4394... Generator Loss: 0.5537
Epoch 1/1... Batch 2940... Discriminator Loss: 1.3943... Generator Loss: 0.5219
Epoch 1/1... Batch 2970... Discriminator Loss: 1.3636... Generator Loss: 0.6176
Epoch 1/1... Batch 3000... Discriminator Loss: 1.2714... Generator Loss: 0.7813
Epoch 1/1... Batch 3030... Discriminator Loss: 1.3552... Generator Loss: 0.7939
Epoch 1/1... Batch 3060... Discriminator Loss: 1.4120... Generator Loss: 0.6244
Epoch 1/1... Batch 3090... Discriminator Loss: 1.3643... Generator Loss: 0.6719
Epoch 1/1... Batch 3120... Discriminator Loss: 1.2840... Generator Loss: 0.9130
Epoch 1/1... Batch 3150... Discriminator Loss: 1.2915... Generator Loss: 0.8690

Submitting This Project

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.